22 Sep Airo Security Announces: ARC: from 3D Game Chips to Licensable RISC Processor
Today marks the 20th anniversary of a landmark in my career, and probably a handful of others. On 21st September 2000, the 32-bit configurable RISC processor company ARC International, now an integral part of Synopsys, completed a major milestone in its history: an IPO (initial public offering) on the London Stock Exchange, at a first day market valuation of around £1.1 billion.
That was a huge achievement for all concerned, but for me it was extra special, because it was the culmination of an objective set by the two people, founder Rick Clucas and CEO Bob Terwilliger, who recruited me as part of the early team of 20 or so people at Argonaut RISC Cores as it was called then, in July 1998. When I visited them for my ‘interview’ in the Argonaut Software building in Edgware, north of London, where there was a large cutout of one the big Nintendo games they’d produced (I believe it was Star Fox) in the lobby, the speed with which startups can make decisions meant I was already offered the ‘job’.
That same evening, pleased that I’d agreed to join, Bob took me to dinner in a well-known Indian restaurant in the center of London, and set me my target: “I want to do an IPO in two years, and you’re going to help us do it,” he told me.
That sounded quite exciting, but I didn’t actually know what it meant. At that point I was quite naïve, as the concept of startups, stock options and IPOs were completely new to me. I had worked for large companies before, like National Semiconductor, Marconi Instruments, and GEC Plessey Semiconductor. I didn’t even know much about RISC processors, apart from meeting Robin Saxby of ARM a few times on my travels to the USA when I was with my previous company. But that was about it.
ARC had an interesting history, which I’d discovered only after I left the company in 2002. It was founded by Rick Clucas and Jez San. Jez is widely known for his success in the games software industry with Argonaut Software, which is where ARC effectively spun out from. Speaking to me last week for this story, Rick said the first form of the ARC processor was actually developed in 1993, and the first licensing deal was with Brooktree in 1996. Prior to the first licensee, the company in several different guises had developed precursors of the ARC processor.
First was the Super FX chip for Nintendo; Jez is said to have received $1 million from Nintendo when they’d turned up in Japan and told them they could develop a hardware accelerator chip that would speed up one of their 3D games by 200 times over the software-based system they were using at the time.
In an interview with Nintendo Life magazine last year, Jez described how Nintendo had a standard DSP chip that didn’t really help them with the graphics so was limited. He said, “We were very forward-thinking with the Super FX chip, because what we designed wasn’t a hardwired polygon engine, it was a fully programmable RISC microprocessor. We had the world’s first GPU years before anyone else had done it, because it wasn’t just the polygons – we could do 3D mapping on it, we could do pixel shading on it… we could do anything, except it only ran on 21MHz.”
After this, Argonaut worked with Hasbro on a chip using three ARC processors for a VR games console, a project which was eventually canned. Next came an assignment for Philips to design a 3D chip for the CD-I 2. And then when that project was canned by Philips, Argonaut designed a 3D chip for a joint venture between Apple and Toshiba for a games console being developed by the Apple Pie division. That also got canned.
Jez said, “We were getting annoyed with these hardware companies hiring us to do really cool hardware and then it didn’t come out, because we were just the engineers working on their product; we didn’t get any say in how it got launched or marketed or anything like that.” But, according to Rick, since the projects were stopped, “we got the rights back.” The processor was actually invented by James Hakewill, with support from Jon Sanders.
So, by the time I joined, there was some history, but by then it had really got to being a proper intellectual property licensing company, through the dedicated hardware spinout, Argonaut RISC Cores.
For me it was a roller-coaster of a journey. I began to understand the culture of a startup before startups were even in vogue. For example, there’s no demarcation of roles. I might have been hired as a director of worldwide communications, but that was just in name only. You did everything: from going out with the CEO and fundraising, to doing the trade shows, organizing graphics, building the exhibition stands, writing press releases, sending out the press releases, building advertising campaigns, writing the brochures and ad copy, hiring people, making the coffee and much more.
That principle didn’t apply just for me — the CEO, the CTO, the engineers, the salespeople, all chip in when you are just 20 or so people trying to build a company. For example, we did many trade shows in the U.S. and Europe. So many that I was traveling to California on a ‘commuting’ basis almost every other week for two years from London. Bob, Rick and I would have competitions on how fast we’d build our exhibition booth or tear it down (it was a pop-up but big enough to require some effort and packing) and get to the airport. Between the three of us and Tim Holden in Europe, David Toombs in the U.S. and Jason Good for both the U.S. and Europe, we’d be in and out of trade shows and conferences, pitching the ARC processor, evangelizing the possibilities of a configurable 32-bit RISC processor and building our partners and networks. And even the engineering team back home got involved in everything — from developing the demos, to helping pack and ship brochures and flyers.
When you do so many things, there are not enough hours in the day. So, I’d often still be in the office at 9:30pm London time talking to the likes of partners and media in California. In my first month, I’d generated headlines in the media like front page stories that said, “ARC challenges ARM,” and I’d spent a week in Sand Hill Road, California, pitching the ARC proposition with Bob and Rick to all the big investors there.
We made a classic mistake when pitching to Silicon Valley investors. We showed them how cool the ARC processor was, with a demo on the laptop showing how anyone could easily configure a processor with whatever instructions you wanted using check boxes on our ‘wizard’ graphic user interface (a name we changed to Architect because another company used ‘wizard’), and then automatically generate the RTL code. The story would have been better received if we talked more about how the benefits of the configurability would have been enough for customers to change their processor.
Despite that ‘failed’ trip to Sand Hill Road, the London-based investors continued to invest in ARC. One of the first people to invest was Tom Teichman, through NewMedia Investors, who’d also invested in Argonaut Games and Lastminute.com. Others early investors included Apax Partners and Nomura, who were able to cash out handsomely on the IPO.
There were some huge challenges as a startup with big-name licensees that couldn’t be named. How could you let the market know? Strict non-disclosure agreements meant you couldn’t use them as reference names for others to have confidence in signing up with you. I remember in 1998 telling the world in our PR and our web site that we had some 18 licensees but could name only three or four of them.
In this respect, credit goes to good reporters in digging for a story. At that time EE Times’ European correspondent Peter Clarke often kept pressing the company to give licensee names, otherwise the story would sound rather hollow (my words not his). It was at the IP 98 show in Frankfurt, Germany, when Peter persisted in asking Bob for names. Peter doesn’t give up in finding the real story, which is one of the characteristics many startups have recognized as a huge benefit when he tells their stories. Next we knew, Peter had published a story after the show deducing Intel was a licensee. That public outage of a licensee didn’t go down too well with Intel, but it was out in public.
As media coverage grew, recruiting became easy — everyone wanted to work for this rising new startup called ARC Cores (the name changed several times, from Argonaut RISC Cores, to ARC Cores, to ARC International pre-IPO). And because it was the dot-com boom…